Estimating People Perception of Intimacy in Daily Life from Context Data Collected with Their Mobile Phone

(1)

Conference Presentation

Reference

WAC, Katarzyna, GUSTARINI, Mattia

Abstract

Nowadays, personal context is extensively used in personalized mobile services. However, gathering this kind of information may compromise the privacy of the user. Moreover, the number of services that collect and share the personal context with others is growing. Yet, the users are not able to easily control what information is shared and with whom. An important step towards this goal is to understand how and when the user wants to share his personal context. The literature suggests that the willingness to disclose personal information depends on the level of intimacy perceived in given contexts. In this paper we show how the MDC data enabled us to estimate the level of intimacy of the user in a given contexts of his/her daily life.

WAC, Katarzyna, GUSTARINI, Mattia. Estimating People Perception of Intimacy in Daily Life from Context Data Collected with Their Mobile Phone. In: Mobile Data Challenge MDC 2012 (by Nokia) Workshop organized in conjunction with the 20th International

Conference on Pervasive Computing 2012, Newcastle (United Kingdom), 18-22 June 2012, 2012, p. 56

Available at:

http://archive-ouverte.unige.ch/unige:72668

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

FACULTY OF ECONOMIC AND SOCIAL SCIENCES

Department of Management Studies

Estimating People Perception of

Intimacy in Daily Life from Context

Data Collected with Their Mobile Phone

Mattia Gustarini, Katarzyna Wac Institute of Services Science

Quality of Life Group

(3)

Introduction

(4)

Introduction

Personal Context

(5)

Introduction

Personal

Context Collected and Shared

Services and

other data collectors

(6)

Introduction

Personal Context

Other People Collected and Shared

Services and

other data collectors

(7)

Introduction

•

Privacy control

•

what information is shared

•

^{with whom}

•

Done by “hand”

•

static preference settings

•

keep them updated

(8)

Goal

privacy sharing

(9)

Goal

privacy sharing

intimacy

(10)

Goal

privacy sharing

intimacy

Help the automation of privacy settings to improve sharing control

(11)

Goal

1. R. S. Gerstein, “Intimacy and privacy,” Ethics, vol. 89, no. 1, pp. 76–81, 1978.

2. J. H. Reiman, “Privacy, intimacy, and personhood,” Philosophy & Public Affairs, vol. 6, no. 1, pp. 26–44, 1976

3. H. T. Reis et al., “Intimacy as an interpersonal process,” in Handbook of personal relationships, vol. 24, no. 3, S. Duck, Ed. Wiley, 1988, pp. 367-389.

privacy sharing

intimacy

(12)

Goal

privacy sharing

intimacy

observers

unwanted degrade intimacy

(13)

Goal

privacy sharing

intimacy

observers

unwanted degrade intimacy

safe places

influence intimacy perception

(14)

Goal

•

Use Mobile Data Challenge data to

estimate participants intimacy level

•

In future it may be possible to use intimacy to help automate privacy

management

(15)

MDC Data Analysis

Observers Safe Places

Bluetooth Charging status Ring status Ring status

Outgoing call/SMS Indoor/Outdoor

(16)

Observers

(17)

Bluetooth

Observers

(18)

Bluetooth

Observers

(19)

stranger

co-worker friend

Bluetooth

Observers

(20)

Ring status

Observers

stranger

co-worker friend

Normal Silent

(21)

Ring status

Observers

stranger

co-worker friend

(22)

Ring status

Observers

stranger

co-worker friend

(23)

Ring status

Observers

stranger

co-worker friend

(24)

Ring status

Observers

stranger

co-worker friend

(25)

Outgoing Call/SMS

Observers

stranger

co-worker friend

(26)

Outgoing Call/SMS

Observers

stranger

co-worker friend

(27)

Outgoing Call/SMS

Observers

call

stranger

co-worker friend

(28)

Outgoing Call/SMS

Observers

call

stranger

co-worker friend

(29)

Outgoing Call/SMS

Observers

stranger

co-worker friend

(30)

Outgoing Call/SMS

Observers

stranger

co-worker friend

sms

(31)

Safe Places

(32)

Charging status

Safe Places

No charging Charging

Park Office

Shop Home

Bus Car

... ...

not safe safe very safe (phone left when full)

(33)

Ring status

Safe Places

Park Office

Shop Home

... ...

not safe safe

external perturbations

call

sms ...

(34)

Indoor / Outdoor

•

Greedy assessment using GPS and WIFI data

•

Outdoor is assumed unsafe

•

Indoor is assumed safe Safe Places

(35)

Intimacy Estimation algorithm

(36)

Intimacy Estimation algorithm

study participation time

(37)

Intimacy Estimation algorithm

study participation time 10 minutes intervals

(38)

Intimacy Estimation algorithm

study participation time 10 minutes intervals

Sun Sat Fr i Thu W ed Tue Mon

01h 03h 05h 07h 09h 11h 13h 15h 17h 19h 21h 23h

1 2 3 4 5 6

₁ ₂ ₃ ₄ ₅ ₆

no intimacy fully intimate intimacy scale

(39)

Intimacy Estimation algorithm

Observers

bluetooth ring status outgoing call/sms

each scores from 1 to 6 average

Safe places

charging status ring status indoor/outdoor

each scores from 1 to 6 average

(40)

•

Intermediate results are

available for

further analysis

•

observers vs safe places

•

single features

Intimacy Estimation algorithm

Observers

Safe places

Intimacy level

average

(41)

Results

Probability

Participants

186 185 179 172 169 165 160 141 139 127 126 123 120 117 111 109 094 089 083 082 077 075 068 063 060 056 051 050 042 034 026 023 017 010 009 007 005 002

0.0 0.2 0.4 0.6 0.8 1.0

1 2 3 4 5 6

(42)

Results

Probability

Participants

186 185 179 172 169 165 160 141 139 127 126 123 120 117 111 109 094 089 083 082 077 075 068 063 060 056 051 050 042 034 026 023 017 010 009 007 005 002

0.0 0.2 0.4 0.6 0.8 1.0

1 2 3 4 5 6 NA

(43)

Results

Probability

Participants

186 185 179 172 169 165 160 141 139 127 126 123 120 117 111 109 094 089 083 082 077 075 068 063 060 056 051 050 042 034 026 023 017 010 009 007 005 002

0.0 0.2 0.4 0.6 0.8 1.0

1 2 3 4 5 6 NA

(44)

Results - P026

Probability

Sun Sat Fri Thu Wed Tue Mon

0.0 0.2 0.4 0.6 0.8 1.0

1 2 34 5 6 NA

Female, 22-27, Student, 21 Weeks (147 days)

(45)

Results - P026

Observers

01h 03h 05h 07h 09h 11h 13h 15h 17h 19h 21h 23h

1 2 3 4 5 6

(46)

Results - P026

Observers

01h 03h 05h 07h 09h 11h 13h 15h 17h 19h 21h 23h

1 2 3 4 5 6

Safe Places

01h 03h 05h 07h 09h 11h 13h 15h 17h 19h 21h 23h

1 2 3 4 5 6

(47)

Results - P026

Intimacy levels

Observers

Safe Places

01h 03h 05h 07h 09h 11h 13h 15h 17h 19h 21h 23h

1 2 3 4 5 6 Sun

Sat Fri Thu Wed Tue Mon

01h 03h 05h 07h 09h 11h 13h 15h 17h 19h 21h 23h

1 2 3 4 5 6

01h 03h 05h 07h 09h 11h 13h 15h 17h 19h 21h 23h

1 2 3 4 5 6

(48)

•

Several limitations on assumptions made for data analysis and intimacy levels

estimation

Limitations

(49)

•

Several limitations on assumptions made for data analysis and intimacy levels

estimation

•

No ground truth available

•

to check the correctness of the approach

•

to verify assumptions

•

to explore more the data

Limitations

(50)

Future work

•

Collect ground truth

•

using Experience Sampling Method

(51)

Future work

•

Verify current assumptions and results

(52)

Future work

•

Explore new data to refine intimacy level estimation

(53)

Future work

•

Explore new data to refine intimacy level estimation

•

Apply automated sharing policies based on intimacy (verify that privacy is respected)

(54)

Conclusions

•

Introduction of the concept of intimacy in privacy management for context sharing

(55)

Conclusions

•

Greedy algorithm to estimate intimacy levels

(56)

Conclusions

•

Greedy algorithm to estimate intimacy levels

•

First results showing that intimacy levels may follow precise patterns

(57)

Thank you!

Mattia Gustarini

mattia.gustarini@unige.ch Katarzyna Wac

katarzyna.wac@unige.ch http://www.qol.unige.ch